气相指纹图谱结合化学计量学分析在中药鉴别中的应用—以3种乳香为例

目的：建立乳香的气相色谱指纹图谱结合化学计量学分析的方法用于3种乳香的鉴别。方法：样品经甲醇超声提取，用HP-5MS毛细管柱采用程序升温的模式进行分离，以FID检测器检测，记录色谱图，并使用ChemPattern软件进行二维聚类分析、主成分分析、相似度分析和模式识别研究。结果：索马里乳香、埃塞俄比亚乳香和印度乳香分别有12、18和25个共有峰。指纹图谱采用二维聚类分析和主成分分析，可区分3种乳香。以23批埃塞俄比亚乳香作为共有模式，3种乳香的相似度均大于0.80，其中所有埃塞俄比亚乳香的相似度均大于0.97。并进一步选择了kNN算法用于模式识别。结论：本方法准确、专属性强、重现性好，可用于乳香的鉴别。

Application of Gas Chromatographic Fingerprint Combined with Chemometrics Analysis in Identification of Traditional Chinese Medicine Taking Three Kinds of Frankincense as Chemometrics

Objective: To establish a gas chromatographic fingerprint combined with chemometrics analysis for the identification of three kinds of frankincense. Methods: The samples were ultrasonically extracted by methanol, separated on the HP-5MS capillary column by a pattern of programmed temperature, and detected by a FID detector. The chromatograms were recorded and 2D cluster analysis, principal component analysis, similarity analysis and pattern recognition study were carried out by using ChemPattern software. Results: There were 12, 18 and 25 common peaks in the chromatograms of Somalis frankincense, Ethiopian frankincense and Indian frankincense respectively. The three kinds of frankincense could be distinguished by 2D cluster analysis and principal component analysis. The similarities of three different kinds of frankincense were greater than 0.80 with the use of 23 batches of Ethiopian frankincense as the common pattern, and the similarities of all Ethiopian frankincense were greater than 0.97. Furthermore, k-nearest neighbor (KNN) algorithm was used for pattern recognition. Conclusion: The method is accurate and shows good specificity and reproducibility, so it can be used for identification of frankincense.